What is the parameter C, in Support Vector Machine Regression?

While working over SVM fundamentally 2 things are considered.

1. Setting a larger margin(To get a better result on test).
2. Keeping Low Misclassification(amount of misclassification of data by model)

But the problem is above both things contradict each other(inversely proportional) i.e if we increase margin its effects misclassification with a high Misclassification rate and if Margin is low which violates our basic requirement №1 but we get low misclassification rate.
But the thing is even we get great low misclassification rate, but the model won’t work properly over validation/unseen data.

So in all this theory where does, parameter C come?

Parameter C represents the size of value margin, higher the value of C Smaller the margin and lower C value means Large Margin.

Larger Margin will cover maximum data points but also it will bring the problem of wrong classifications.

Smaller Margin Will satisfy lower wrong classification but will end up with false prediction over unseen data.

You can relate this issue with the problem of Model Over-fitting and Under-fitting.

So in short C value must be given carefully because it will decide your model’s behaviour and correctness over unseen data.

So what is the best value for C, well it’s all up on your model and honestly it’s all about trial and error (cross-validation), change the value and crosscheck the results.

Please, Correct if am wrong or suggest me any additional points to it.

References:
https://medium.com/@pushkarmandot/what-is-the-significance-of-c-value-in-support-vector-machine-28224e852c5a

PS: This post is taken from my own medium post.:
https://medium.com/@rathodrutesh/what-is-parameter-c-in-support-vector-machine-regression-f75be9df98ec